42
Table 8: The Effect of a Valence Advantage on U.S. Senate Incumbents’
Margins of Victory and Likelihoods of Winning, 1996-2002
Independent Variables
Model 1 (Linear):
Dep. Var.: Margin of
Victory (Loss) in General
Election (%)
Coefficient (s.e.)
Model 2 (Logit):
Dep. Var.: Won = 1;
0 = Lost
Coefficient (s.e.)
Valence Advantage:
Project Outlays/State Popu.
7.097 (3.251)**
0.721 (1.045)
Deviation of Senator From
Constituency Median
(Absolute Value)
0.629 (5.046)
-2.937 (1.100)***
Freshman Senator
-3.741 (3.393)
-0.952 (0.714)
Quality Challenger in
General Election
-20.221 (3.630)***
-1.385 (0.738)*
State Vote for Presidential
Candidate of
Incumbent’s Party (%)
0.848 (0.221)***
0.079 (0.049)
Constant
-22.934 (12.925)*
-0.167 (1.045)
Wald χ
2
74.58***
17.83***
N
†
103
104
*p<0.10; **p<0.05; ***p<0.01; 2-tailed tests. Both equations are estimated with random-effects for each
state.
†Model 2 includes Senator Bob Smith (R-NH) who was defeated in the Republican primary, while he is
obviously not included in Model 1 since the dependent variable is the general election margin.
Convention